auto generated data 2

[Dominique Michel <dominique michel vtxnet ch>]

Hi,

Forget my preceding message, this one is much better:

I am a beginner with gnumeric, and want to use data generated from data
sheets curves by engauge-digitizer in gnumeric. They are in .csv files
that look like this:

x,-35,-30,-25,-20,-15,-10,-5,0,5,10,15,20,25,30
0,0,0,0,0,0,0,0,0,0,0,0,0,0,0
0.703889,,,,,0.00141043,,,,,,,,,
0.705549,0.000706213,,0.000706213,,,,0.000706213,,,,,0.000706213,,
0.707209,,1.99214e-06,,,,,,1.99214e-06,,,,,1.99214e-06,1.99214e-06
1.40944,,,,0.00211665,,,,,,,,,,
1.41276,,,,,,,,,,,0.000708205,,,
2.10503,,,,,0.00704818,,,,,,,,,
2.10835,,,,,,,,,,0.00563974,,,,
2.11333,,,,,,0.00352708,,,,,,,,
2.81389,,0.00634595,,,,,0.00634595,,,,,,,
2.81721,,,,,,,,,,,,,0.00493751,
2.81887,,,,,,,,,0.00423329,,,,,
3.51612,,,,,,,,,,,,0.00846061,,
3.51944,,,,0.00705217,,,,,,,,,,
3.5211,,,,,,,,,,,,,,0.00634795
3.52276,,,,,,,,,,,0.00564373,,,
3.5294,0.00282684,,,,,,,,,,,,,
...
770.122,0.0141412,,,,,,,,,,,,,

The first column is the x axes, and each subsequent column represent
the y values of one digitized curve on a curves set.
On the first line, the values (beside the x) represent Ug1 for
the column, so we have different curves at different Ug1.

The only column that have a value for each line is the first one, x.
But I also have 2 other sets of curves, where the curves are not always
beginning at the first line (at x=0), so it can be missing values also
at the beginning.

What I want to do is to use gnumeric to transform those curves
(constant voltage vacuum tubes characteristics) into another set of
curves (constant current characteristics), and from there, calculate the
parameters for the corresponding spice tube model.

Based on those constant current curves, a spice model that take in
account the tube saturation become much more easier to develop.
I need this in order to model correctly stuffs like guitar amp.

The tube in use is a pentode with 5 connections: cathode, g1, g2, g3
and anode. To measure it, the cathode and g3 is at the ground, g2
is at a fixed and constant voltage. That give 3 sets of curves when
measured in this constant voltage mode:

Ia=f(Ua)) for different Ug1
Ig2=f(Ua) for different Ug1
Ig1=f(Ua) for different Ug1
So x is Ua, and y is Ia, Ig2 or Ig1.

I want to transform them into at least
Ug1=f(Ua) for different Ia
Ug1=f(Ua) for different Ig2
and eventually
Ug1=f(Ua) for different Ig1
x become Ug1, and y become Ua

I can use engauge to extrapolate the missing values, but as anyway,
I must extrapolate the values for the new sets of curves, it look to be
simpler to only use gnumeric for that.

What I have to do for a given new curve, as example
Ug1=f(Ua); Ia=0.01;0.05;0.1;0.5
is to search for the values of Ug1 and Ua that correspond at those
different Ia.
Note also than, at a given Ia, not all the original
curves will have value (for Ia=200mA, only the curves with Ug1>=0 will
have values.

That imply several extrapolations. 
- For a given Ia, I must search for the nearest values in the
  original sheet, calculate the Ua values that correspond to the
  given Ia, and get the corresponding Ug1.
- As the values of Ug1 are very few, it would also be great if I can
  combine this first step and extrapolate (Ug1,Ua,Ia) values for
  extrapolated Ug1.

Excuse me to be so long, but English is not my native language, and I
want to be as sure as possible than you understand my problem.

I think than, as engauge take a lot of points, something like
y=(x*y1)/x1 will be enough for the extrapolations. At least for the
first step without extra Ug1 values.

My first concern is how to search for the nearest values of some given
number into several columns with a lot of blank cells. How can I do
that?

Cheers,
Dominique